What the published figures say about artisanal cobalt.

No census exists. Figures split on geography (whole Copperbelt vs Lualaba-only) and on whether they count diggers or everyone at a site (one mapping reaches ~430,000 counting all roles). But the deepest problem is provenance: the only real head-counts (Öko-Institut, BGR) land near 67,000–110,000, while the widely quoted 150k–255k figures are either untraceable round numbers or back-calculations from a child-labour ratio. The most quoted numbers are the least measured.

Every figure here is two estimates stacked: a soft miner count multiplied by an assumed family size. The multiplier alone ranges from 4× to 6×. Nobody counts dependents directly.

There is only one number, and it is often quoted for more than it measures. The "40,000 children mining cobalt" repeated in press and lawsuits is a single 2014 UNICEF estimate for children in all artisanal mining across southern DRC: all minerals. No measured count of children mining cobalt specifically has ever been made.

These numbers are barely comparable. Some are net cash in hand, others gross; they come from different sites and cooperatives; and above all they straddle the cobalt price cycle: the same hour of digging paid several times more at the 2018 peak than in the 2024 slump. Within a single study the spread runs from under a dollar to over fifty.

This is a documented blank. The only solid anchor (40%) is all-mineral, all-roles. Women in cobalt mostly wash, sort and carry, and seldom dig, so they fall out of "digger" counts entirely, and no one has published a cobalt-specific share.

A thirty-fold gap, explained by method. The World Bank counts only press-reported deaths (a known floor); Kara scales up from questionnaires (a disputed ceiling). No mortality register exists, and every source agrees the real toll is undercounted.

Most of this spread is real change over time: the DRC's share climbed from ~64% in the mid-2010s to the mid-70s as new industrial mines came online, then eased to about 73% in 2025 as Indonesia grew. The rest is rounding (the IEA's 70% vs USGS's 76%). A widely circulated '84%' figure is an aggregator error inconsistent with its own USGS source, and is not used here.

This is the most contested number on the page, and the best lesson in why a figure needs a year stamped on it. Artisanal supply is extremely price-elastic: ~20–40% at the 2018 peak, under 2% by 2024. The widely quoted "20%" (Amnesty, 2016) and "12%" (RAID, 2022) are pre-collapse figures still circulating as if current. ASM cobalt is also informal and mixed into industrial streams, so every reading is an estimate.

CTCPM’s 2025 administrative total (100,015 t) is far below the USGS 2025 mine-production estimate (230,000 t). Their coverage and measurement basis have not been reconciled, so neither is substituted for the other. The other readings also mix years, estimates and projections; world totals must not be read as DRC totals.

Two things move this: periodic USGS revisions (DRC reserves at ~6 Mt; the world total lifted from ~11 Mt to ~12 Mt in 2026, nudging the share from ~55% toward ~50%), and the reserves-versus-resources confusion. World cobalt resources run to ~25 Mt: a far larger denominator than reserves: so "half of reserves" and "half of resources" are different claims often blurred, and DRC production share (~73%) is a third, distinct number.

This is the rare figure that IS measured, traded daily on the LME, and it is still useless as a fixed fact. It roughly halved and doubled within five years. Watch two traps: 2018 and 2022 were both "peaks" (different years, different highs), and metal, hydroxide and LME-cash are distinct price series quoted interchangeably. Even a measured price goes out of date.

These four numbers measure four different things: government tax take, gross export value, and total world market size: that get conflated constantly. Price volatility then swings any tonnage's value two- or three-fold year to year. And the artisanal slice has no value series at all: ASM cobalt is informal and mixed into industrial exports, so nobody prices it separately.

The gap here is mostly definitional: "EVs" (43%) are a subset of "all batteries" (76%), and the battery share is genuinely rising year to year. The classic trap sits just off this chart: the USGS "51% superalloys" figure is US-only consumption, constantly mis-quoted as the global picture, where batteries dominate.